Shaking Up (and Keeping Intact) the Old Boys’ Network: The Impact of the Mandatory Gender Quota on the Board of Directors in India

The term “caste” in India can refer to two distinct concepts—the varna and the jati (Deshpande, 2011). Several prior studies on caste have looked at the broader classification of varna (Ajit et al., 2012; Bapuji and Chrispal, 2020; Dayanandan et al., 2019), where Brahmins, Kshatriyas, and Vaishyas are deemed “upper castes” or “forward castes” while Shudras, and “untouchables” are deemed “lower castes”. However, for official purposes, a legal classification is used which comprises of three broad categories: (a) the former “untouchable” castes (Dalits), who are collectively classified as Scheduled Castes (SC), (b) socially and economically marginalized indigenous ethnic groups that are classified as Scheduled Tribes (ST) and (c) other “lower castes” as Other Backward Classes (OBC). Although, there is no official estimate of Forward Caste share in the population, a rough estimate can be arrived at by subtracting the cumulative share of OBC, SC and ST from 100 percent. For our empirical analysis, similar to Ajit et al. (2012), we operationalize caste of directors using this legal classification. Together we term SC, ST and OBC castes as underprivileged castes because separating them further results in such low numbers that there is no variation left to run the regression. Subtracting the cumulative share of OBC, SC, ST directors gives us an estimate of “forward caste” which we term privileged castes (Damaraju & Makhija, 2018; Munshi, 2019).

For the purposes of this study, we classify directors into trading and non-trading communities. Following prior research on Indian businesses (Damodaran 2008; Mani 2019; Mani & Durand, 2019; Timberg, 1978), we identify 14 prominent trading communities in India, where trading communities are “groups that have an economic focus ingrained in their culture and belief systems” (Mani & Durand, 2019).

Director status is operationalized as eigenvector centrality which measures the level of connectedness based on a director’s connection with other highly connected directors.

Securities and Exchange Board of India, the regulator overseeing listed companies and stock exchanges in India, also revised its Equity Listing Agreement to align with the requirements of the 2013 act, and required all listed companies to appoint at least one woman director to their boards no later than April 1, 2015 (Deloitte, 2013).

The law is applicable to every listed company (except those having paid up equity share capital not exceeding 10 crore and net worth not exceeding 25 crore) and any public company having a paid-up share capital of 100 crore or more, or, a turnover of 300 crore (Vohra, 2020).

Aroras and Khatris are Sikh trading castes. However, prior research categorizes them as Hindu privileged castes because of the similarity between Hindu and Sikh religious orders (Berreman, 1972). For example, Risley (1892) writes: “If then, it is at all necessary to connect the Khatris with the ancient fourfold system of castes, the only group to which we can affiliate them is the Vaisyas”.

Consistent with prior studies (Vohra, 2020), we observe a high rate of compliance with the law: The number of firms without a single woman director decreased significantly (65% in 2013 to less than 1% in 2018) and the number of women directors increased significantly (5% in 2013 to 15% in 2018). In addition, a significant proportion of firms (25% in 2018) went beyond the minimum requirement of having one woman director.

This official list of underprivileged castes is a systematic pan-India list but has the limitation that it lists caste names (not last names associated with each caste). We therefore validate this caste classification by comparing it with two alternate classifications. First, for 240 of the most frequently occurring last names in our data (accounting for 49.83% of our overall data), we identified the castes manually, starting with matrimonial sites, and when unavailable, supplementing with other public sources (for example, dedicated caste websites, ancestry sources, reports, and articles in the business press, social media pages of different castes and communities, and mobile applications such as India Caste Hub). Second, we created a last name to caste mapping based on the merit list of OBC, SC, and ST students released by AICTE (All India Counsel for Technical Education). The overall matches between the classifications using these two alternate mappings and our primary classification using the official OBC, SC, and ST lists are 94% and 90%, respectively. We also rerun our regression analysis using these two alternate mappings and our conclusions remain the same. We thank the editor for pointing us to these alternative methods of mapping last names to caste names.

‘Gujarati’ is an identity based on region and includes many religions and castes. For robustness, we also rerun our analysis after recategorizing Gujaratis as a “non-trading community”, and after dropping them from the analysis. Our conclusions do not significantly change in either specification. Thanks to the editor for pointing out this issue.

Results of robustness tests available upon request.